Built independently by an author, for readers. Read the story and support ChapterPal
ChapterPal Logo
Log inSign up

keyword

back-translation quality

Back-translation quality refers to the linguistic fluency, semantic accuracy, and fidelity of synthetic text generated when an automated translation model translates target-language text back into a source language. In machine translation and natural language processing workflows, back-translation serves as a data augmentation technique to convert abundant target-side monolingual text into paired bilingual training data. The quality of these generated source sentences determines the overall utility of the augmented dataset, where fluent and semantically faithful translations effectively regularize downstream models and broaden vocabulary coverage, while low-quality translations risk introducing misleading syntactic noise, hallucinated content, or translation errors that can degrade model performance.

1 item

Improving Neural Machine Translation Models with Monolingual Data

Improving Neural Machine Translation Models with Monolingual Data

Rico Sennrich, Barry Haddow, Alexandra Birch

OrganizationsUniversity of Edinburgh

Why you should read this

Introduces back-translation to train neural machine translation models on target-side monolingual data without architecture changes, substantially increasing translation accuracy across standard and low-resource benchmarks.

Neural Machine Translation (NMT) has obtained state-of-the art performance for several language pairs, while only using parallel data for training. Target-side monolingual data plays an important role in boosting fluency for phrase-based statistical machine translation, and we investigate the use of monolingual data for NMT. In contrast to previous work, which combines NMT models with separately trained language models, we note that encoder-decoder NMT architectures already have the capacity to learn the same information as a language model, and we explore strategies to train with monolingual data without changing the neural network architecture. By pairing monolingual training data with an automatic back-translation, we can treat it as additional parallel training data, and we obtain substantial improvements on the WMT 15 task English<->German (+2.8-3.7 BLEU), and for the low-resourced IWSLT 14 task Turkish->English (+2.1-3.4 BLEU), obtaining new state-of-the-art results. We also show that fine-tuning on in-domain monolingual and parallel data gives substantial improvements for the IWSLT 15 task English->German.

Added

2026-09-13